January	
  2017	
  
Industrial	
  Analy.cs	
  2016	
  /	
  2017	
  
	
  
	
  
	
  
	
  
Frank	
  Pörschmann	
  
	
  
Frank.poerschmann@digital-­‐analy=cs-­‐associa=on.de	
  
2	
  
•  The	
  focus	
  is	
  on	
  the	
  promo.on	
  of	
  data	
  competency	
  in	
  business,	
  poli=cs	
  and	
  
society.	
  
•  For	
  more	
  than	
  10	
  years,	
  the	
  Digital	
  Analy=cs	
  Associa=on	
  (DAA),	
  with	
  over	
  5,000	
  
members	
  worldwide,	
  has	
  been	
  suppor=ng	
  the	
  professionaliza=on	
  of	
  the	
  new,	
  
data-­‐driven	
  professional	
  images,	
  such	
  as	
  the	
  digital	
  analyst	
  and	
  the	
  data	
  scien=st.	
  
•  The	
  Digital	
  Analy=cs	
  Associa=on	
  e.V.	
  is	
  con=nuing	
  this	
  commitment	
  as	
  an	
  
independent	
  non-­‐profit	
  organiza.on	
  under	
  its	
  own	
  na=onal	
  leadership.	
  
•  The	
  Digital	
  Analy=cs	
  Associa=on	
  e.V.	
  supports	
  ins.tu.ons,	
  specialists	
  and	
  
execu.ves	
  in	
  the	
  development	
  of	
  professional	
  and	
  entrepreneurial	
  skills	
  for	
  the	
  
analysis	
  of	
  digital	
  data	
  streams.	
  
3	
  
§  Qualifica=on	
  &	
  Cer=fica=on	
  
§  Promo=on	
  of	
  Young	
  
§  Networking	
  
§  Events	
  
§  Research	
  &	
  Development	
  
§  Wegweiser	
  für	
  Unternehmen	
  und	
  
Anwender	
  
§  Career	
  development	
  and	
  support	
  	
  
§  Advisory	
  services(i.e.	
  data	
  rights,	
  project	
  
management,	
  tooling)	
  
	
  
§  Science	
  &	
  Educa=on	
  ||	
  Promo=on	
  of	
  Young	
  
§  Business	
  &	
  Governance	
  	
  
§  Soware	
  Producer	
  ||	
  Agencies	
  &	
  Service	
  
Companies	
  
§  Methods	
  ||	
  Knowledge	
  Management	
  
§  Interna=onal	
  ||	
  Networking	
  
§  Marke=ng,	
  PR	
  &	
  Events	
  ||	
  Members	
  
§  Legal	
  
Ac.vi.es	
   Subject	
  MaMer	
  Expert	
  Groups	
  
„Professionaliza-on	
  of	
  data-­‐driven	
  professions	
  	
  
for	
  data	
  expert	
  as	
  well	
  as	
  management.	
  “	
  
Global	
  representa-ve	
  decision-­‐maker	
  study	
  
	
  available	
  for	
  download	
  	
  
www.IoT-­‐Analy=cs.de	
  
	
  
www.digital-­‐analy=cs-­‐associa=on.de	
  
The	
  concept	
  of	
  digital	
  shadow	
  applies	
  to	
  machines	
  as	
  well	
  
The	
  most	
  expecisve	
  cost	
  factor	
  in	
  business	
  s.ll	
  	
  
are	
  bad	
  decisions	
  
Signals	
  
Data	
  
Informa=on	
  
Decision	
  
Knowledge	
  
Gathering/management/Distribu=on	
  	
  
Analy=cs	
  
Advanced	
  Analy=cs	
  &	
  Data	
  Science	
  
-­‐  Learning	
  systems,	
  ML,	
  AI	
  
-­‐  Decision	
  Support	
  Systems	
  
-­‐  Automated	
  decision	
  support	
  systems	
  	
  
Conversion	
  /	
  Distribu=on	
  	
  
Repor=ng	
  /	
  Monitoring	
  
o 	
  Daten-­‐Analy=cs	
  is	
  not	
  new	
  –	
  but	
  different	
  	
  
	
  
o  60‘S	
  &	
  70‘s:	
  	
  	
  	
  	
  	
  	
  	
  Opera=onal	
  Monitoring	
  	
  
o  2017:	
   	
  Decision	
  Support	
  
o  2025:	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
   	
  Automated	
  Decision-­‐making	
  	
  
Industrial	
  Analy-cs...	
  
o is	
  a	
  key	
  success	
  factor	
  of	
  Industrial	
  
Internet	
  (Industrie	
  4.0)	
  
o Will	
  become	
  a	
  compe==ve	
  cri=cal	
  
capability	
  in	
  industrial	
  business	
  	
  
o For	
  most	
  of	
  the	
  companies	
  integrated	
  
data	
  analy=cs	
  is	
  a	
  new	
  organiza=onal	
  
discipline	
  (approx.	
  ½	
  within	
  one	
  org.-­‐
unit,	
  mainly	
  R&D)	
  	
  
o BUT:	
  	
  
30%	
  report	
  about	
  finalized	
  projects	
  
Development	
  of	
  data	
  competencies	
  already	
  on	
  	
  
top-­‐management	
  agenda	
  	
  
o More	
  than	
  half	
  ini=ated	
  by	
  CEO	
  &	
  COO	
  
Data	
  is	
  not	
  IT!	
  
	
  
o Smallest	
  responsibility	
  by	
  typical	
  
technology	
  management	
  CTO/CIO	
  
33,1%	
  
Value	
  expecta-ons	
  on	
  industrial	
  analy-cs	
  mainly	
  set	
  on	
  
growth	
  instead	
  of	
  efficiency	
  
o Expected	
  benefit	
  from	
  analy=cs	
  mainly	
  
in	
  growth	
  and	
  customer	
  sa=sfac=on	
  
o Expected	
  growth	
  by:	
  
o  Extending	
  exis=ng	
  products	
  
o  Expansion	
  of	
  exis=ng	
  business	
  models	
  
o  New	
  Business	
  Models	
  
o Cost-­‐reduc=on	
  and	
  efficiency	
  increase?	
  
Rela=vely	
  weakly	
  weighted	
  despite	
  
numerous	
  successful	
  projects	
  
33,1%	
  
How	
  good	
  are	
  you	
  at..?	
  
o Over	
  half	
  of	
  the	
  companies	
  are	
  
sa=sfied	
  with	
  the	
  ability	
  to	
  access	
  their	
  
data	
  
o But	
  about	
  2/3	
  of	
  the	
  companies	
  fail	
  in	
  
genera=ng	
  sufficiently	
  relevant	
  
findings	
  from	
  the	
  data	
  obtained.	
  
	
  
-­‐>	
  But	
  this	
  is	
  the	
  source	
  of	
  future	
  
	
  	
  	
  	
  	
  compe==ve	
  advantage	
  	
  
33,1%	
  
Further:	
  Cost	
  and	
  benefit	
  structures	
  are	
  currently	
  unbalanced	
  
o Main	
  costs	
  of	
  data	
  projects	
  in	
  IT	
  &	
  
Technology	
  disciplines	
  
o Business	
  benefit	
  is	
  generated	
  by	
  
analy=cs.	
  Analy=cs	
  costs	
  account	
  for	
  
only	
  about	
  25%	
  of	
  the	
  total	
  costs	
  
o Strategies	
  for	
  cost-­‐saving	
  data	
  
architectures	
  are	
  already	
  relevant	
  
€	
  
S-ll	
  one	
  Use	
  Case	
  seem	
  to	
  prevail:	
  Predic-ve	
  Maintenance	
  PdM)	
  
Comparison	
  of	
  maintenance	
  approaches	
  
Deep-­‐dive:	
  Condi=on	
  Based	
  Maintenance	
  vs.	
  Predic=ve	
  Maintenance	
  
Copyright	
  ©	
  2016	
  by	
  www.iot-­‐analy=cs.com	
  All	
  rights	
  reserved	
  
13	
  
Condi.on-­‐based	
  maintenance	
   Predic.ve	
  Maintenance	
  
Mobile	
  Condi-­‐	
  
.on	
  Monitoring	
  
Online	
  Condi.on	
  
Monitoring	
  
Sta.s.cs-­‐based	
  	
  
PdM	
  
Stochas.c-­‐based	
  	
  
PdM*	
  
Sensing	
  
Technology	
  
Handheld	
  Device	
   Integrated	
  Sensors	
   Integrated	
  Sensors	
   Integrated	
  Sensors	
  
Monitoring	
  
Frequency	
  
In	
  regular	
  intervals	
  /	
  on	
  
demand	
  
Constantly	
   Constantly	
   Constantly	
  
Visualiza.on	
   On	
  specific	
  device	
   Online	
  /	
  Mobile	
   Online	
  /	
  Mobile	
   Online	
  /	
  Mobile	
  
IT-­‐Architecture	
   On-­‐Premise	
   On-­‐Premise	
  or	
  Cloud	
   On-­‐Premise	
  or	
  Cloud	
   On-­‐Premise	
  or	
  Cloud	
  
Real-­‐.me	
  
monitoring	
  
Combina.on	
  of	
  
data	
  sources	
  
Analy.cs	
   Sta=s=cs1	
   Stochas=cs2	
  
Maintenance	
  
Trigger	
  
If	
  monitoring	
  shows	
  cri=cal	
  
values	
  
If	
  monitoring	
  shows	
  cri=cal	
  
values	
  
When	
  calculated	
  health-­‐
score	
  reaches	
  cri=cal	
  value	
  
When	
  failure	
  is	
  predicted	
  to	
  
occur	
  
a.	
   b.	
   a.	
   b.	
  
*	
  =	
  Some=mes	
  also	
  referred	
  to	
  as	
  “prognos=cs”	
  	
  1.	
  Sta=s=cs	
  =	
  Using	
  sta=s=cal	
  methods	
  such	
  as	
  SPSS,	
  regression	
  	
  2.	
  Stochas=cs	
  =	
  Using	
  stochas=cal	
  models	
  such	
  as	
  Bayesian	
  Networks,	
  etc.	
  
Most	
  challenging	
  issues	
  
o Security	
  remains	
  strong	
  obstacle	
  
o However,	
  the	
  biggest	
  hurdles:	
  
o  Interoperability	
  of	
  systems	
  
o  Quality	
  of	
  the	
  data	
  
o  Insight	
  genera=on	
  by	
  lack	
  of	
  	
  
specialists,	
  skills,	
  methods,	
  	
  
tools…	
  
33,1%	
  
A	
  rapid	
  shiK	
  of	
  tools	
  requires	
  new	
  skills	
  and	
  capabili-es	
  
o The	
  end	
  of	
  spreadsheet	
  analy=cs	
  
o Rapid	
  change	
  of	
  tools	
  and	
  
playorms	
  to	
  be	
  experienced	
  
within	
  5	
  years	
  
o Importance	
  of	
  predic=ve	
  analy=cs	
  
tools	
  will	
  double	
  
o BI	
  relevance	
  increases	
  as	
  well	
  
	
  
33,1%	
  
Which	
  approach	
  to	
  use?	
  Freestyle	
  or	
  structured?	
  
o About	
  2/3	
  work	
  on	
  hypothesis	
  
	
  from	
  the	
  begin	
  
o S=ll	
  1/3	
  allows	
  for	
  gaining	
  insight	
  
in	
  their	
  own	
  data	
  
33,1%	
  
Cri-cal	
  skill	
  gap	
  ahead	
  
	
  
Warning	
  
Data	
  Science	
  
	
  -­‐	
  Data	
  Scien=st,	
  Data	
  Engineers	
  	
  
-­‐  IT	
  (Developer,	
  Architects,	
  SI,	
  
Infrastructur	
  (M2M)	
  
-­‐  Agile	
  PM	
  
-­‐  Industrial	
  process	
  know-­‐how	
  
	
  
Companies	
  fail	
  in	
  integra=ng	
  adequate	
  
new	
  digital	
  professions	
  	
  
	
  
Only	
  5	
  years	
  to	
  go	
  un=l	
  skill	
  gap	
  
impacts	
  compe==ve	
  capability	
  
33,1%	
  
Digital	
  sovereignty	
  ...?	
  
o Promo=on	
  of	
  data	
  competency	
  among	
  
specialists	
  and	
  execu=ves	
  is	
  crucial	
  for	
  
Europe’s	
  industrial	
  strategy	
  
o The	
  German	
  educa=on	
  system	
  in	
  new	
  
data-­‐driven	
  professions	
  is	
  
interna=onally	
  not	
  compe==ve.	
  
o Companies	
  must	
  take	
  on-­‐the-­‐job	
  
qualifica=on	
  and	
  interna=onalize	
  skills.	
  
33,1%	
  HandcraKs	
  have	
  many	
  faces,	
  	
  
so	
  does	
  data	
  art.	
  
19	
  
Contact	
  
	
  
Frank	
  Pörschmann	
  
Crémon	
  36	
  
20457	
  Hamburg	
  
	
  
0171	
  –	
  30	
  579	
  20	
  
	
  
Frank.Poerschmann@digital-­‐analy=cs-­‐associa=on.de	
  
	
  
www.didital-­‐analy=cs-­‐associa=on.de	
  
www.digitalanaly=csassocia=on.org	
  
	
  

Industrial Analytics and Predictive Maintenance 2017 - 2022

  • 1.
    January  2017   Industrial  Analy.cs  2016  /  2017           Frank  Pörschmann     Frank.poerschmann@digital-­‐analy=cs-­‐associa=on.de  
  • 2.
    2   •  The  focus  is  on  the  promo.on  of  data  competency  in  business,  poli=cs  and   society.   •  For  more  than  10  years,  the  Digital  Analy=cs  Associa=on  (DAA),  with  over  5,000   members  worldwide,  has  been  suppor=ng  the  professionaliza=on  of  the  new,   data-­‐driven  professional  images,  such  as  the  digital  analyst  and  the  data  scien=st.   •  The  Digital  Analy=cs  Associa=on  e.V.  is  con=nuing  this  commitment  as  an   independent  non-­‐profit  organiza.on  under  its  own  na=onal  leadership.   •  The  Digital  Analy=cs  Associa=on  e.V.  supports  ins.tu.ons,  specialists  and   execu.ves  in  the  development  of  professional  and  entrepreneurial  skills  for  the   analysis  of  digital  data  streams.  
  • 3.
    3   §  Qualifica=on  &  Cer=fica=on   §  Promo=on  of  Young   §  Networking   §  Events   §  Research  &  Development   §  Wegweiser  für  Unternehmen  und   Anwender   §  Career  development  and  support     §  Advisory  services(i.e.  data  rights,  project   management,  tooling)     §  Science  &  Educa=on  ||  Promo=on  of  Young   §  Business  &  Governance     §  Soware  Producer  ||  Agencies  &  Service   Companies   §  Methods  ||  Knowledge  Management   §  Interna=onal  ||  Networking   §  Marke=ng,  PR  &  Events  ||  Members   §  Legal   Ac.vi.es   Subject  MaMer  Expert  Groups   „Professionaliza-on  of  data-­‐driven  professions     for  data  expert  as  well  as  management.  “  
  • 4.
    Global  representa-ve  decision-­‐maker  study    available  for  download     www.IoT-­‐Analy=cs.de     www.digital-­‐analy=cs-­‐associa=on.de  
  • 5.
    The  concept  of  digital  shadow  applies  to  machines  as  well  
  • 6.
    The  most  expecisve  cost  factor  in  business  s.ll     are  bad  decisions   Signals   Data   Informa=on   Decision   Knowledge   Gathering/management/Distribu=on     Analy=cs   Advanced  Analy=cs  &  Data  Science   -­‐  Learning  systems,  ML,  AI   -­‐  Decision  Support  Systems   -­‐  Automated  decision  support  systems     Conversion  /  Distribu=on     Repor=ng  /  Monitoring   o   Daten-­‐Analy=cs  is  not  new  –  but  different       o  60‘S  &  70‘s:                Opera=onal  Monitoring     o  2017:    Decision  Support   o  2025:                        Automated  Decision-­‐making    
  • 7.
    Industrial  Analy-cs...   o is  a  key  success  factor  of  Industrial   Internet  (Industrie  4.0)   o Will  become  a  compe==ve  cri=cal   capability  in  industrial  business     o For  most  of  the  companies  integrated   data  analy=cs  is  a  new  organiza=onal   discipline  (approx.  ½  within  one  org.-­‐ unit,  mainly  R&D)     o BUT:     30%  report  about  finalized  projects  
  • 8.
    Development  of  data  competencies  already  on     top-­‐management  agenda     o More  than  half  ini=ated  by  CEO  &  COO   Data  is  not  IT!     o Smallest  responsibility  by  typical   technology  management  CTO/CIO   33,1%  
  • 9.
    Value  expecta-ons  on  industrial  analy-cs  mainly  set  on   growth  instead  of  efficiency   o Expected  benefit  from  analy=cs  mainly   in  growth  and  customer  sa=sfac=on   o Expected  growth  by:   o  Extending  exis=ng  products   o  Expansion  of  exis=ng  business  models   o  New  Business  Models   o Cost-­‐reduc=on  and  efficiency  increase?   Rela=vely  weakly  weighted  despite   numerous  successful  projects   33,1%  
  • 10.
    How  good  are  you  at..?   o Over  half  of  the  companies  are   sa=sfied  with  the  ability  to  access  their   data   o But  about  2/3  of  the  companies  fail  in   genera=ng  sufficiently  relevant   findings  from  the  data  obtained.     -­‐>  But  this  is  the  source  of  future            compe==ve  advantage     33,1%  
  • 11.
    Further:  Cost  and  benefit  structures  are  currently  unbalanced   o Main  costs  of  data  projects  in  IT  &   Technology  disciplines   o Business  benefit  is  generated  by   analy=cs.  Analy=cs  costs  account  for   only  about  25%  of  the  total  costs   o Strategies  for  cost-­‐saving  data   architectures  are  already  relevant   €  
  • 12.
    S-ll  one  Use  Case  seem  to  prevail:  Predic-ve  Maintenance  PdM)  
  • 13.
    Comparison  of  maintenance  approaches   Deep-­‐dive:  Condi=on  Based  Maintenance  vs.  Predic=ve  Maintenance   Copyright  ©  2016  by  www.iot-­‐analy=cs.com  All  rights  reserved   13   Condi.on-­‐based  maintenance   Predic.ve  Maintenance   Mobile  Condi-­‐   .on  Monitoring   Online  Condi.on   Monitoring   Sta.s.cs-­‐based     PdM   Stochas.c-­‐based     PdM*   Sensing   Technology   Handheld  Device   Integrated  Sensors   Integrated  Sensors   Integrated  Sensors   Monitoring   Frequency   In  regular  intervals  /  on   demand   Constantly   Constantly   Constantly   Visualiza.on   On  specific  device   Online  /  Mobile   Online  /  Mobile   Online  /  Mobile   IT-­‐Architecture   On-­‐Premise   On-­‐Premise  or  Cloud   On-­‐Premise  or  Cloud   On-­‐Premise  or  Cloud   Real-­‐.me   monitoring   Combina.on  of   data  sources   Analy.cs   Sta=s=cs1   Stochas=cs2   Maintenance   Trigger   If  monitoring  shows  cri=cal   values   If  monitoring  shows  cri=cal   values   When  calculated  health-­‐ score  reaches  cri=cal  value   When  failure  is  predicted  to   occur   a.   b.   a.   b.   *  =  Some=mes  also  referred  to  as  “prognos=cs”    1.  Sta=s=cs  =  Using  sta=s=cal  methods  such  as  SPSS,  regression    2.  Stochas=cs  =  Using  stochas=cal  models  such  as  Bayesian  Networks,  etc.  
  • 14.
    Most  challenging  issues   o Security  remains  strong  obstacle   o However,  the  biggest  hurdles:   o  Interoperability  of  systems   o  Quality  of  the  data   o  Insight  genera=on  by  lack  of     specialists,  skills,  methods,     tools…   33,1%  
  • 15.
    A  rapid  shiK  of  tools  requires  new  skills  and  capabili-es   o The  end  of  spreadsheet  analy=cs   o Rapid  change  of  tools  and   playorms  to  be  experienced   within  5  years   o Importance  of  predic=ve  analy=cs   tools  will  double   o BI  relevance  increases  as  well     33,1%  
  • 16.
    Which  approach  to  use?  Freestyle  or  structured?   o About  2/3  work  on  hypothesis    from  the  begin   o S=ll  1/3  allows  for  gaining  insight   in  their  own  data   33,1%  
  • 17.
    Cri-cal  skill  gap  ahead     Warning   Data  Science    -­‐  Data  Scien=st,  Data  Engineers     -­‐  IT  (Developer,  Architects,  SI,   Infrastructur  (M2M)   -­‐  Agile  PM   -­‐  Industrial  process  know-­‐how     Companies  fail  in  integra=ng  adequate   new  digital  professions       Only  5  years  to  go  un=l  skill  gap   impacts  compe==ve  capability   33,1%  
  • 18.
    Digital  sovereignty  ...?   o Promo=on  of  data  competency  among   specialists  and  execu=ves  is  crucial  for   Europe’s  industrial  strategy   o The  German  educa=on  system  in  new   data-­‐driven  professions  is   interna=onally  not  compe==ve.   o Companies  must  take  on-­‐the-­‐job   qualifica=on  and  interna=onalize  skills.   33,1%  HandcraKs  have  many  faces,     so  does  data  art.  
  • 19.
    19   Contact     Frank  Pörschmann   Crémon  36   20457  Hamburg     0171  –  30  579  20     Frank.Poerschmann@digital-­‐analy=cs-­‐associa=on.de     www.didital-­‐analy=cs-­‐associa=on.de   www.digitalanaly=csassocia=on.org